Analyses

Hierarchical regression

Add predictors in blocks and see how much R² changes.

Edit on GitHub

When to use it

Use hierarchical regression when you, not an algorithm, decide the order. For example: enter study hours first, then ask whether anxiety still explains exam score after hours are already in the model.

Assumptions

Same straight-line model as linear regression. Tensr does not add residual plots or a collinearity table on this procedure.

Running it in Tensr

Analyze → Regression → Linear → Hierarchical. In chat: “Hierarchical regression of score, hours in block 1, hours and anxiety in block 2.”
Dependent is the outcome. Blocks is a list of predictor lists, at least one block.
Method is enter only. Confidence level starts at 0.95.

Options

Prop

Type

Reading the output

Hours go in first. Anxiety is added in a second block. Each block was built to add a slice of R², not to swallow the outcome.

BlockPredictorsR²Adj R²ΔR²F-changep-change
1hours0.1070.097———
2anxiety0.2310.2150.12415.047< .001
VariableBSEβ95% CIp-value
hours2.0620.6190.303[0.832, 3.292].001
anxiety-2.0920.539-0.353[-3.162, -1.021]< .001

Hierarchical regression for score with 2 block(s). Hierarchical regression for score with 2 block(s). Metrics: Final R² = 0.231; Adj R² = 0.215; N = 96.

Reporting (APA 7)

Hierarchical regression for score with 2 block(s). Report the estimate in the table. This procedure is not summarised by one p-value.

Coming from SPSS

Analyze → Regression → Linear → Hierarchical.

SPSS hierarchical regression is the Block button in Linear Regression: you assign predictors to Block 1 of 1, then the next block. Tensr takes those blocks as a list.

Letting the software pick the order is the Stepwise regression section of linear regression. A single block is that same page with method enter.